This semester I am teaching a seminar on computational syntax. It’s mostly on subregular syntax, but I started out with a discussion of CCG. CCG is noteworthy because it is a theory-rich approach that has managed to make major inroads into NLP. It would be cool if we could replicate this with MGs, but in order to do that you need a killer app. Subregular complexity might just be that because CCG doesn’t have a regular backbone, so it can’t have a subregular one either (more on that in a future post). CCG’s killer app was flexible constituency and a one-to-one mapping from syntax to semantics. You combine that with a corpus (CCGbank) and an efficient parsing algorithm (e.g. supertagging with A* parsing), and you have something that is both linguistically sophisticated and sufficiently fast and robust for practical applications. Anyways, this post collects some of my thoughts on flexible constituency and how it could be emulated in MGs. Spoiler: shells, lots and lots of shells.

Continue reading

As you might know, Stony Brook hosted AMP, the American Meeting on Phonology, a week ago quite a while ago (yikes, almost November again). Jane Chandlee started off the conference with an invited talk on the subregular view of phonological mappings from underlying representations to surface forms. It was well received, but during the question period Bruce Hayes (no, not that Bruce Hayes) made a point that I found puzzling: “You need a framework!” Unfortunately I didn’t have time to ask Bruce afterwards what exactly he meant by that. But every conceivable interpretation I’ve come up with I vehemently disagree with, and I think Bruce’s demand for a framework stems from incorrectly applying the linguistic modus operandi to computational work.

Continue reading

Oh boy, this month is killing me. I know I promised you one more detailed discussion of subregular complexity before we finally get back to the topic that started it all, Omer Preminger’s post on listedness. But in the interest of time I’ll merge those two posts into one. That means some points will be a bit more hazy than I’d like, but I think it will work just fine nevertheless (and for those of you sick of this series of posts, we’ll get to something new sooner). Alright, with that out of the way, here’s the basic question: why aren’t PF and LF more alike?

Continue reading

The previous post in this series discussed the lay of the (is)land from the perspective of TSL (I’m so disappointed in myself for not making this pun last time; better late than never). I mentioned that the TSL view cannot handle all island constraints. Sometimes, we need an alternative approach. But this alternative approach doesn’t come out of nowhere. It is also what we need for all kinds of licensing constraints, and it also handles restrictions on movement that are not island constraints.

Continue reading

In the previous post we saw Merge is SL-2 over dependency trees, and Move is TSL-2. For every movement feature f we project a separate tier that contains only lexical items that have a licensor feature f⁺ or a licensee feature f^-. A tier is well-formed iff every lexical item with a licensee feature has a mother with a licensor feature, and every lexical item with a licensor feature has exactly one lexical item among its daughters that carries a licensee feature. It’s a pretty simple system. Despite that simplicity, it predicts a fundamental aspect of movement: island effects!

Continue reading

Alright, syntax. Things are gonna get a bit more… convoluted? Nah, interesting! In principle we’ll see a lot of the same things as in phonology, and that’s kind of the point: phonology and syntax are actually very similar. But syntax isn’t quite as exhibitionist as phonology, it doesn’t show off its subregular complexity in the open for the world to see. So the first thing we’ll need is a suitable representation. Once we have that, it’s pretty much phonology all over again, but now with trees.

Continue reading

After a brief interlude, let’s get back to locality. This post will largely act as a recap of what has come before and provide a segue from phonology to syntax. That’s also a good time to look at the bigger picture, which goes beyond putting various phenomena in various locality boxes just because we can.

Continue reading

Pro tip: Don’t start a multi-part series of posts on locality right before the beginning of the semester and when you have a pile of papers to review. On the upside, this will give you guys some extra time to digest all the concepts in the three previous posts. In the meantime, here’s a quick and dirty post on corpus linguists and why it should be part of our syntax curriculum. I didn’t even proofread it, so beware.

Continue reading

We’re continuing our stroll through the subregular locality zoo, still in the service of ultimately being able to say something about syntax and its interface with PF and LF. We started out with SL as the most restrictive kind of locality. The step to the TSL region corresponds to a relativized notion of locality where it’s not absolute locality that matters, but your position relative to other elements of a specific type. TSL is actually a cluster of different classes, with standard TSL at the very bottom. Depending on what context information the TSL tier projection may take into account, TSL expands to ITSL, OTSL, or IOTSL. While IOTSL has a very liberal notion of locality that can even accommodate insane patterns like nati, it still involves some level of locality. That’s in stark contrast to the strictly piecewise (SP) dependencies of Rogers, Heinz, Bailey, Edlefsen, Vischer, Wellcome, and Wibel (2010), which throw locality out the window.

Continue reading

The previous post covered the essentials of strictly local (SL) and tier-based strictly local (TSL) dependencies over strings. We saw that even though TSL generalizes SL to a relativized notion of locality, it is still a restrictive model in the sense that not every non-local dependency is TSL. In principle that’s a nice thing, but among those non-local dependencies beyond the purview of TSL we also find some robustly attested phenomena like unbounded tone plateauing. Fair enough, but that does not mean that unbounded tone plateauing is entirely non-local.

Continue reading